We report here on a comparison of three models, Model of the Global Universal Tracer Transport in the Atmosphere (MOGUNTIA, abbreviated here as MOG), European Center/Hamburg (GCM Model) 3 (ECHAM3, here E3), and European Center/Hamburg (GCM Model) 4 (ECHAM4, here E4), with a decade-long time series of sulfate aerosol measurements (Obs) from the Mauna Loa Observatory (MLO), Hawaii. The observations were sorted with four criteria to eliminate any contamination from the local volcanoes. E3 was run for 8 years, and E4 was run for 5 years to assess interannual variability. In terms of the climatological average concentration, E3 (51 pmol mol-1) was very close to Obs (53 pmol mol-1), while MOG (33 pmol mol-1) and E4 (134 pmol mol-1) predicted different concentrations. The interannual variability of the annual average was somewhat larger in Obs than in E3 and E4. The large interannual variability is noteworthy; it demonstrates that multiyear time series are essential for comparison with climatological models. The seasonal cycle of Obs was reproduced surprisingly well by E3. E4 predicted much higher concentrations, although the seasonal amplitude was reasonable. MOG showed much less seasonal variability, which suggests that stochastic processes (which are not described by climatological models) may be responsible for much of the transport from continents to remote regions. Using tagged sources, E4 found that urban/industrial sources in SE Asia and volcanoes were responsible for most of the sulfate at MLO. The interannual variability of each month was similar for Obs, E3, and E4. When E4 was run in a nudged mode for 1 year (so that it tended to reproduce the actual meteorology) and compared day-by-day to Obs, the R2 was 0.73. Thus, in spite of significant magnitude differences, E4 clearly reproduces many of the features that cause free-tropospheric sulfate concentrations to vary from day-to-day. When we moved the continual 5--8 km volcanic source in E4 to lower altitudes, the large overestimate was reduced but not eliminated, suggesting that the high-altitude releases alone cannot explain the high concentrations predicted by E4. Missing processes and the parameterizations of vertical exchange and scavenging are among the possible reasons for the differences between the simulations. While the discrepancy between the observations and E4 suggests that we first look for shortcomings in E4's sulfur processing, we cannot exclude the possibility that compensating errors in E3 cause its better agreement with the observations. ¿ 2001 American Geophysical Union